blind-spot-detective by jwynia/agent-skills
npx skills add https://github.com/jwynia/agent-skills --skill blind-spot-detective系统地识别非虚构写作中缺失的内容——包括盲点(你方法固有的局限性)和空白点(可以弥补的缺口)。提供用于发现遗漏、检验假设和确保全面覆盖的框架。
你看不到的东西比你能看到的更重要。 识别缺失的内容比认识到已包含的内容更困难。系统性的审视能揭示随意审阅所遗漏的缺口。
| 类型 | 定义 | 解决方案 |
|---|---|---|
| 盲点 | 你的方法论、理论或视角固有的局限性 | 调整你的方法或承认局限性 |
| 空白点 | 在你当前方法框架内可以弥补的缺口 | 在现有框架内扩展覆盖范围 |
关键见解: 了解你正在处理哪种类型的缺口,决定了是应该改变你的方法还是仅仅扩展它。
处理支持现有信念的信息,同时忽略矛盾的信息。
自我检查:
假设读者拥有与你相同的专业知识。
自我检查:
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
能识别他人的偏见,却看不到自己的。
自我检查:
| 问题类型 | 目的 | 示例 |
|---|---|---|
| 澄清 | 探索复杂想法 | 我到底是什么意思?这与我的主要论点有何关联? |
| 假设探究 | 揭示隐藏的假设 | 我把什么视为理所当然?这背后有哪些未言明的信念? |
| 证据与推理 | 评估支持论据的质量 | 有什么证据支持这一点?证据充分吗?我的结论是否成立? |
| 替代观点 | 挑战默认框架 | 不同学科会如何看待这个问题?批评者会怎么说? |
假设你以以下身份阅读:
| 缺口类型 | 措施 |
|---|---|
| 缺失证据 | 添加来源、示例或数据 |
| 缺失视角 | 寻求该群体的意见或承认缺口 |
| 缺失逻辑 | 添加明确的推理或过渡 |
| 缺失背景 | 添加背景信息或定义 |
| 固有局限性 | 在范围声明中予以承认 |
模式: 利用盲点检测来拖延完成。每发现一个缺口就导致更多分析。永远没有足够完整的时候。失败原因: 写作永远不可能完美。盲点检测是为了识别重大的遗漏,而不是达到不可能的完整性。修正: 设定阈值。“我将处理那些从根本上削弱我论点的缺口,而不是所有可能的扩展。” 为检测过程设定时间限制。
模式: 在拥有草稿之前就进行盲点分析。在尚不存在的东西中寻找缺口。失败原因: 盲点检测适用于已存在的写作。你需要有东西来分析。草稿前的缺口焦虑会阻碍你开始。修正: 先写,后检测。先完成一个完整的草稿,然后识别缺失的内容。在具体的文本中比在抽象的计划中更容易看到缺口。
模式: 将每个识别出的缺口都视为需要处理的事项。不断扩展范围,直到作品变得难以管理。失败原因: 并非每个缺口都需要填补。有些缺口在范围内是合适的。试图处理所有事情会产生臃肿、不集中的写作。修正: 区分盲点(承认局限性)和空白点(扩展覆盖范围)。对于每个缺口,问:“填补这个对我的论点是否至关重要?”
模式: 认为你需要外部专家来识别盲点。等待外部验证,而不是进行系统的自我审查。失败原因: 虽然外部视角有帮助,但你可以通过系统性的框架自己识别许多盲点。依赖会造成瓶颈。修正: 先使用这些框架。获取外部审查是为了验证,而不是为了发现。大多数明显的缺口可以通过结构化的自我审视找到。
模式: 生成详尽的缺失元素清单,但没有优先考虑哪些最重要。失败原因: 一份包含 50 个项目的缺口清单会让人瘫痪。并非所有缺口都同等重要。没有优先级,作者要么放弃,要么随机处理缺口。修正: 按严重性分类:关键(削弱论点)、重要(削弱论证)、次要(可以增强)。首先处理关键的。
输入:
输出:
互补技能:
non-fiction-revision: 用于实施修正summarization: 用于测试论点清晰度research: 用于填补证据缺口每周安装数
110
代码仓库
GitHub 星标数
42
首次出现
Jan 20, 2026
安全审计
安装于
opencode91
gemini-cli86
codex86
claude-code81
github-copilot79
cursor77
Systematically identify what's missing in non-fiction writing—both blind spots (inherent limitations of your approach) and blank spots (gaps that could be addressed). Provides frameworks for finding omissions, testing assumptions, and ensuring comprehensive coverage.
What you can't see matters more than what you can. Identifying what's missing is harder than recognizing what's included. Systematic interrogation reveals gaps that casual review misses.
| Type | Definition | Solution |
|---|---|---|
| Blind Spots | Limitations inherent to your methodology, theory, or perspective | Adjust your approach or acknowledge limitations |
| Blank Spots | Gaps that could be addressed within your current approach | Expand coverage within existing framework |
Key insight: Understanding which type of gap you're dealing with determines whether to change your approach or simply expand it.
Processing information that confirms existing beliefs while ignoring contradictions.
Self-check:
Assuming readers share your specialized knowledge.
Self-check:
Recognizing biases in others but not yourself.
Self-check:
| Question Type | Purpose | Examples |
|---|---|---|
| Clarification | Explore complex ideas | What exactly do I mean? How does this relate to my main argument? |
| Assumption-Probing | Uncover hidden assumptions | What am I taking for granted? What unstated beliefs underlie this? |
| Evidence & Reasoning | Evaluate support quality | What evidence supports this? Is it sufficient? Does my conclusion follow? |
| Alternative Viewpoints | Challenge default framework | How would a different discipline view this? What would critics say? |
Read as if you:
| Gap Type | Action |
|---|---|
| Missing evidence | Add sources, examples, or data |
| Missing perspective | Seek input from that group or acknowledge gap |
| Missing logic | Add explicit reasoning or transitions |
| Missing context | Add background or definitions |
| Inherent limitation | Acknowledge in scope statement |
Pattern: Using blind spot detection to delay finishing. Every gap found leads to more analysis. Nothing is ever complete enough. Why it fails: Writing is never perfect. Blind spot detection is for identifying significant omissions, not achieving impossible completeness. Fix: Set a threshold. "I will address gaps that fundamentally undermine my argument, not every possible expansion." Time-box the detection process.
Pattern: Running blind spot analysis before having a draft. Looking for gaps in something that doesn't exist yet. Why it fails: Blind spot detection works on existing writing. You need something to analyze. Pre-draft gap anxiety prevents ever starting. Fix: Write first, detect second. Get a complete draft, then identify what's missing. Gaps are easier to see in concrete text than abstract plans.
Pattern: Treating every identified gap as something to address. Expanding scope until the piece becomes unmanageable. Why it fails: Not every gap needs filling. Some gaps are appropriate for scope. Trying to address everything produces bloated, unfocused writing. Fix: Distinguish blind spots (acknowledge limitation) from blank spots (expand coverage). For each gap, ask: "Is filling this essential to my thesis?"
Pattern: Believing you need outside experts to identify blind spots. Waiting for external validation instead of systematic self-review. Why it fails: While outside perspectives help, you can identify many blind spots yourself with systematic frameworks. Dependency creates bottlenecks. Fix: Use the frameworks first. Get external review for validation, not discovery. Most obvious gaps can be found with structured self-interrogation.
Pattern: Producing exhaustive lists of missing elements without prioritizing which matter most. Why it fails: A 50-item gap list is paralyzing. Not all gaps are equal. Without priority, writers either give up or address gaps randomly. Fix: Categorize by severity: critical (undermines thesis), significant (weakens argument), minor (would enhance). Address critical first.
Inbound:
Outbound:
Complementary:
non-fiction-revision: For implementing fixessummarization: For testing thesis clarityresearch: For filling evidence gapsWeekly Installs
110
Repository
GitHub Stars
42
First Seen
Jan 20, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
opencode91
gemini-cli86
codex86
claude-code81
github-copilot79
cursor77
冲刺回顾模板:敏捷团队回顾会议指南与模板(开始-停止-继续/愤怒-悲伤-高兴/4Ls)
10,400 周安装